Probability of detection analysis of round robin test results performed by flash thermography

Within the scope of a standardisation research project, a flash thermography round robin test that evaluates reliability, comparability and efficiency of different testing situations was performed. Data recorded at metal test specimens with flat bottom holes (FBHs) were analysed by calculating the signal-to-noise ratio (SNR) of the defect signatures in the thermograms as well as in the phase images as a function of defect parameters. A new multi-parameter probability of detection (POD) model was developed, where an a versus a continuous signal analysis was based on the linear relationship between the SNR and a multi-parameter a. This linear relationship was verified by comparison to data obtained from an analytical model that is considering lateral thermal heat diffusion as well as to data obtained by numerical simulation. The resulting POD curves for the thermograms and phase images give an estimation for the detectability of the FBHs with known geometry in steel using different equipment and obtained by different participants. By comparing the SNRs of FBHs with similar geometries, this POD model was transferred to aluminium and copper as well.